**** BEGIN LOGGING AT Fri Mar 24 10:59:56 2006 Mar 24 12:18:51 Hi all. Is dyoung around? After his last commit I get errors in bitbake: ImportError: cannot import name methodpool Mar 24 12:19:07 anyone that knows a fix? Mar 24 12:27:05 I fixed that again. Mar 24 12:27:24 rm -rm bitbake; make setup Mar 24 12:29:06 still the same Mar 24 12:30:31 Hah youre right. Mar 24 12:30:37 Just moment Mar 24 12:30:54 (fixed buy forgot to push it) Mar 24 12:31:36 :) Mar 24 12:34:42 okay give that a try. Mar 24 12:36:00 thx. it works now Mar 24 12:36:05 np. Mar 24 12:37:12 has anyone detected any improvements in the clock stability with the 2.6.16 kernel? Mar 24 12:40:14 the clock should be more or less correct with 2.6.16 Mar 24 12:41:19 My network is still running on 2.6.12 and we are experiencing a lot of strange issues with the clock stability and ntp Mar 24 12:42:00 I am considering an upgrade at the moment, hoping that some of the issues might be resolved in the newer kernels Mar 24 12:42:33 hanjo: what is your tickadj set to ? Mar 24 12:42:52 I think it is the default Mar 24 12:43:42 we get errors like: x1205_sync_rtc exit (change seconds 2) Mar 24 12:44:07 ntpd or openntpd? Mar 24 12:45:27 ntp proper Mar 24 12:45:31 ntpd Mar 24 12:46:40 isn't tickadj depreciated for Linux? Mar 24 12:51:44 is your clock swinging widely? Mar 24 12:52:15 because theres a 33.000 Mhz crystal on it, but its supposed to be a 33.333Mhz Mar 24 12:52:21 the funny thing is that it is happening only on some slugs, while others are fine Mar 24 12:52:52 and 33333333/33000000 is around a tickadj of 10101 Mar 24 12:54:11 has this been made the default? Mar 24 12:54:42 it was hacked into the kernel at some point; but I dont remember when. Mar 24 12:54:49 sec Mar 24 12:54:57 in openslug 2.7-beta, it wasn't in the kernel Mar 24 12:55:05 in 3.6-beta, there's an in-kernel fix Mar 24 12:56:06 ok, just run a test Mar 24 12:56:24 all my slugs are tick = 10101 Mar 24 13:04:01 since you guys are here, can someone also fix slugos-packages Mar 24 13:04:46 it depends on slugos-image and slugos-native, but these packages provide ${SLUGOS_IMAGENAME}-native and ${SLUGOS_IMAGENAME}-image Mar 24 13:06:03 actually, slugos-native should probably just become build-essentials Mar 24 13:06:13 I don't have my devbox handy thoughj Mar 24 14:00:20 grmbl Mar 24 14:00:29 suddenly my 2.5" disk refuses to spin up Mar 24 14:00:40 * NAiL goes *grrrr* Mar 24 14:08:02 NAiL, they're keep you pretty busy. :) Mar 24 14:08:28 s/keep/keeping/ Mar 24 14:08:29 VoodooZ meant: NAiL, they're keeping you pretty busy. :) Mar 24 14:13:33 Why are both UHCI, EHCI and OHCI installed on the slug? The Slug only has an EHCI and OHCI?? Mar 24 14:14:59 I never remember the difference there Mar 24 14:15:21 UHCI is mostly on x86s PC I believe. Mar 24 14:15:40 OHCI/UHCI are similar I believe. Mar 24 14:18:08 then I guess UHCI should be removed ;) Mar 24 14:18:31 I'm still struggling to wake up, so I'm not gonna try fixing much stuff yet. Mar 24 14:19:01 well, I'm not sure anymore as it seems every year somebody comes back and adds it in. Unless people are just unsure. Mar 24 14:19:26 NAiL, Don't worry about it. It's a minor thing. Concentrate on the release. Mar 24 14:19:31 hehe Mar 24 14:19:34 yeah, I should Mar 24 14:19:45 Thanks for your great work btw. Mar 24 14:19:49 mostly the script missing still Mar 24 14:19:58 I wish I could help out more but that robot is driving mad. Mar 24 14:20:07 the robot is mad? ;) Mar 24 14:20:07 And on top of that I'm in the process of changing job! Mar 24 14:20:10 hehehe Mar 24 14:20:15 yep! Mar 24 14:20:56 We'll get around to the release soon Mar 24 14:21:09 seems like I'm more or less the only one working on it though Mar 24 14:25:40 I think I'm gonna try take a nap and see if I can get my head back into shape Mar 24 14:29:57 yep. a nap solves a lot of problems! Mar 24 15:01:14 What are the remaining issues with the new release? I've just tried baking a new image from the latest monotone pull, but it makes the slug not accessible over ssh Mar 24 15:06:36 the release sources is in SVN, not monotone Mar 24 15:06:43 s/is/are/ Mar 24 15:06:44 NAiL meant: the release sources are in SVN, not monotone Mar 24 15:07:07 the outstanding issues are basically the leds shellscript and testing. Mar 24 15:08:05 so should monotone work? Mar 24 15:09:35 dunno, I'm trying not to do much work with the monotone repo and instead focusing on the release Mar 24 15:12:13 Is the monotone repo being kept in sync with the SVN or you are hacking on the SVN until everythin runs smoothly and than push to monotone? Mar 24 15:13:18 no, they're currently not being kept in sync Mar 24 15:13:42 the svn repo is intended to be stable Mar 24 15:13:53 whereas the monotone repo has a lot of changes Mar 24 15:14:25 There are some fixes that I'm doing to SVN that I'm going to push to monotone afterwards, but I'm putting that off so I'm not going to get confused. Mar 24 15:15:15 (Yes, that's kinda backwards. Development is supposed to happen in monotone then get pushed to svn if it works. But the two repos are very different right now) Mar 24 15:22:05 I see. Thanks. Keep up the good work. We can do some testing if you like. We have enough spare slugs. Mar 24 15:32:21 That would be great, once the leds shellscript is written Mar 24 15:35:42 OK. Mar 24 15:36:20 bbl Mar 24 15:37:55 NAiL: does the current head build for you? I get errors in libc: ./bits/endian.h:13:2: #error Machine byte order unknown. Mar 24 15:38:04 no Mar 24 15:42:50 marceln, ping Mar 24 16:13:48 03bzhou * 10unslung/ (make/swi-prolog.mk sources/swi-prolog/swi-prolog.mk): changed to use swipl as PL name, handles x86_64 x-build machine now, may break native build Mar 24 16:15:17 Hi eFfeM Mar 24 16:24:08 hi can anyone help me here with SLug and SAS Mar 24 16:24:11 L Mar 24 16:30:02 Hi eFfeM Mar 24 16:30:16 NAiL: ping? Mar 24 16:30:33 eFfeM: i solved the usb module problem Mar 24 16:30:52 not ran the image yet, but at least modules are now copied to rootfs Mar 24 16:32:22 all: who should be contacted for slugbug ? Mar 24 16:33:06 i tried creating an account but never received mail with password Mar 24 16:33:19 mailserver was not even contacted for trying to deliver mail Mar 24 16:33:51 pepijn: pong Mar 24 16:34:00 hei Mar 24 16:34:31 the change you made yesterday evening did not work for EfFeM and me because we run debianslug, not openslug ;) Mar 24 16:34:36 try rwhitby/ka6sox for the slugbug issues. They'll probably figure out ;) Mar 24 16:34:47 ah, I haven't done the change to all yet Mar 24 16:34:50 added the lines to debianslug.conf instead Mar 24 16:34:57 That's on my to-do list ;) Mar 24 16:35:29 Currently working on the leds script Mar 24 16:35:49 bu then: modules are not called ehci, ohci, uhci but all have -hcd at the end Mar 24 16:35:54 right now it parses most of the arguments, and is able to switch on and off a led :-P Mar 24 16:36:06 so correct lines are: kernel-module-ehci-hcd \ Mar 24 16:36:06 kernel-module-ohci-hcd \ Mar 24 16:36:06 kernel-module-uhci-hcd \ Mar 24 16:36:06 kernel-module-sd-mod \ Mar 24 16:36:06 kernel-module-usb-storage \ Mar 24 16:36:07 kernel-module-scsi-mod \ Mar 24 16:36:13 aha, will fix Mar 24 16:36:16 plan to test run the image build with that soon Mar 24 16:36:35 hehe, ref. leds Mar 24 16:36:39 a miracle! Mar 24 16:36:42 ;) Mar 24 16:43:33 gaah Mar 24 16:43:39 this trigger stuff is making me nuts Mar 24 16:43:48 there's so many possibilities Mar 24 16:44:24 eg, the cpu trigger has idle/busy which again has on/off/flash Mar 24 16:45:12 "leds status cpu idle on" currently makes the amber light go on when the cpu is idle. Mar 24 16:45:14 yeah, It did look overly convoluted Mar 24 16:46:10 I'm going to rip all this off anyways as I need the status LED as my robot's heartbeat LED. Mar 24 16:46:44 heartbeat? Mar 24 16:47:09 leds disk-1 timer 250 250 Mar 24 16:47:12 :-P Mar 24 16:49:26 A simple flashing LED that indicates the main sensor loop is working (toggling) Mar 24 16:50:36 * VoodooZ goes out for lunch Mar 24 16:50:38 ah Mar 24 16:56:36 hmmm slug is not responding on the network now Mar 24 16:57:51 been away for dinner Mar 24 16:58:03 pepijn i am not using debianslug, I am on openslug Mar 24 16:59:17 the reason it did not work for me was because i use the mastermakefile and apparently the changed slugos.conf didn't make it to nslu2-linux yet Mar 24 16:59:51 the slugbug issue was already mentioned on the mailing list, it seems to be a disk full error Mar 24 16:59:57 at least that is what I recall Mar 24 17:03:23 NAiL: the modules listed by pepijn, are they supposed to be in my openslug.conf file? I just did a make update but it is not in my openslug.conf Mar 24 17:08:25 * ByronT-Away is now auto-away after 3h idle Mar 24 17:10:34 eFfeM: they haven't gone into monotone yet. Mar 24 17:11:25 ooooooh, i saw a cia msg yesterday, so I expected this to be here by now Mar 24 17:11:40 eFfeM: http://svn.berlios.de/viewcvs/openslug/releases/slugos-3.6-beta/openembedded/conf/distro/openslug.conf?rev=291&r1=257&r2=291 Mar 24 17:12:32 ok understood, tnx, cant test this now anyway as glibc is giving compiler errors Mar 24 17:13:01 * NAiL is busy working on the leds-script. I'm just checking in once in a while ;) Mar 24 17:13:36 NAiL: np, I'll wait it out Mar 24 17:13:45 unless of course there is anything I can do to help Mar 24 17:14:40 the odd thing is that bitbake starts giving me errors as well and I didn't touch it Mar 24 17:14:41 NOTE: Using cache in '/home/frans/OpenSlugHead/openslug/tmp/cache' Mar 24 17:14:41 ERROR: 'str' object has no attribute 'keys' while parsing /home/frans/OpenSlugHead/openslug/openembedded/packages/alsa/alsa-driver_0.9.6-hh4c.bb Mar 24 17:14:59 and that for about 10 files Mar 24 17:15:09 feels like the ground is crumbling away below my feet ... Mar 24 17:16:43 remove the cache Mar 24 17:16:54 (I think) ;) Mar 24 17:17:02 there was a mail on the OE ml aabout it Mar 24 17:17:05 s/aabout/about/ Mar 24 17:17:06 NAiL meant: there was a mail on the OE ml about it Mar 24 17:33:29 anyon up for testing a basic leds script? Mar 24 17:33:53 It's still lacking errorchecking stuff Mar 24 17:34:00 hi there, dwery :) Mar 24 17:34:07 wanna test the leds script? ;) Mar 24 17:34:15 scirpt? Mar 24 17:34:20 yup Mar 24 17:34:31 for controlling the leds using the new interface Mar 24 17:34:37 (the old binary doesn't work) Mar 24 17:35:12 let me see it Mar 24 17:35:19 http://osx.kynisk.com/leds Mar 24 17:35:36 Don't complain about my shellscript-fu. I know it's not good :-P Mar 24 17:38:48 :D Mar 24 17:39:08 NAiL: tnx, will try Mar 24 17:39:36 need some errorchecking on timer and cpu triggers, but it seems to mostly work now Mar 24 17:39:57 seems fine. you need to use 255 instead of 1 for led on Mar 24 17:40:07 ah, yes Mar 24 17:40:10 not all the triggers might be available but i don't think there's a way to lnow Mar 24 17:40:30 no, I know... Mar 24 17:40:52 that'll take a bit more work. But the way it is now is probably good enough for the slug ;) Mar 24 17:41:30 I agree Mar 24 17:41:33 And this is more or less the last piece needed before release Mar 24 17:41:36 NAiL: removed the cache and indeed the problem is gone. I'm not on the oe mailing list, so I was unaware of it Mar 24 17:42:00 actually this also solved my libc compilation problem Mar 24 17:42:05 good :) Mar 24 17:43:34 Ok, I'll be going off for a while to fix a monitor. Please try to break my script, and give me feedback. I'll be reading the log when I get back. Mar 24 17:45:55 ok Mar 24 17:51:45 (updated script to list available triggers) Mar 24 17:52:31 :-D Mar 24 17:53:05 (and fixed the formatting) Mar 24 17:53:40 ugh Mar 24 17:53:47 gerbleh Mar 24 17:53:53 that makes it wrong :-P Mar 24 17:54:31 there we are Mar 24 17:55:03 I split up the cpu trigger into cpu-idle and cpu-busy, since it makes it easier to program Mar 24 17:59:15 methinks the release will be ready for testing later tonight Mar 24 18:00:27 Yipee! Mar 24 18:00:42 I'll be testing it! Mar 24 18:00:46 good :) Mar 24 18:01:32 Although I should be saving my old 2.7 image as I have too much work left to do to fight with openslug. Just in case. Mar 24 18:12:31 ok, so the known issues that have to be fixed are having the proper modules in the rootfs, add the leds script and fix the initscripts so they use the correct leds stuff. Then the release should be (hopefully) ok. Mar 24 18:12:54 I'll be back in a couple of hours. Mar 24 18:18:30 AwayNAiL, did you catch a nap after all? Mar 24 19:03:49 NAiL: just flashed slugos-3.6-beta / debianslug-image and although the modules are now there, they are not loaded Mar 24 19:04:26 also, the slug appeared on 192.168.1.77 and with 'default' SysConf Mar 24 19:07:57 ok, default sysconf might be leftover from last night ;) Mar 24 19:08:53 03bzhou * 10unslung/make/less.mk: upstream upgrade to 394 Mar 24 19:10:30 yes, it was related to a cleanout of SysConf and not running 'turnup preserve' Mar 24 19:10:55 but the not-loading of the modules still is an issue Mar 24 19:11:04 Hi there - i guess my question is so easy, that it is not answered in any Place i searched... Mar 24 19:13:46 ...so here it is: I got Unslung5.5 running on my Slug and a MemoryStick in USB1. Now i'd like to unsling to a second Disk to be able to run two different Systems. Can i just boot the Slug withount any disk connectend and then connect my USB-Drive to Port 1 and unsling again? Or will that stop my Flash-Drive workin when i conect it to Port 1 next Week? Mar 24 19:15:08 Oh, maybe i am on the wron channel - sorry! Mar 24 19:18:07 MarvinGS: you're not necessarily on the wrong channel :) Mar 24 19:18:17 Glad you asked before you did it. Mar 24 19:18:32 mwester: ah, ok :) Mar 24 19:18:49 so i guess i would have messed up everything ;) Mar 24 19:20:15 An "unsling" is a one-time operation as far as the NSLU2 is concerned. So if you unsling a second time, the new disk you unslung to would have a partial filesystem. You'll need to reflash the NSLU2 with a new copy of the Unslung 5.5 firmware, then unsling to the disk. At that point you can do what you are trying to to: it will boot up equally well on either disk (just use the same USB... Mar 24 19:20:16 ...Port for the disk and flash disk). Mar 24 19:22:56 Ah, ok. So no Disk-Geometry or "type" (Flashdrive/Disk) or something is stored in the Slug when unslinging - just the Number of the USB-Port. Good to know that. Is anything else i could do after unslinging stored inside the Slug? Like the downloaded ipkg stuff? Mar 24 19:31:03 Once you've booted up in "unslung" mode, the flash filesystem internal to the NSLU2 is completely out of the picture -- so you can have completely independent installs on the two devices. For example, I have one tiny hard drive that has a build environment installed -- so if I ever need to do a native compile, I can unplug the big disk, and boot it up on a separate device for my development... Mar 24 19:31:04 ...efforts. It's really rather neat, although I'm not sure that it was an intended feature! :) Mar 24 19:35:31 snowbike: anything? Mar 24 19:39:01 how can I reconfugre the kernel build for opendebianslug? Mar 24 19:43:40 s/builld/built/ Mar 24 20:01:33 Sorry, i pulled the wrong plug ;) Mar 24 20:01:55 mwester: can you please repeat the last text? Mar 24 20:09:14 MarvinGS: Once you've booted up in "unslung" mode, the flash filesystem internal to the NSLU2 is completely out of the picture -- so you can have completely independent installs on the two devices. For example, I have one tiny hard drive that has a build environment installed -- so if I ever need to do a native compile, I can unplug the big disk, and boot it up on a separate device for my... Mar 24 20:09:16 ...development efforts. It's really rather neat, although I'm not sure that it was an intended feature! :) Mar 24 20:09:30 Ok, thank you! Mar 24 20:11:43 Do you know any place with information about what happens when you unsling to a "fresh" disk? I'd like to add a FAQ Entry for this topic. Mar 24 20:14:18 I'd challenge you to pick a spot. Where should it have been on the wiki for you to have found it? If anyone objects with where you create/add the info, they'll move it themselves. Mar 24 20:14:45 I wonder why some FAQ Entry even suggest you to unsling a second time... Mar 24 20:14:49 OK, i'll do that! Mar 24 20:16:37 I found that suggestion at the end of http://www.nslu2-linux.org/wiki/FAQ/DiskNotRecognized Mar 24 20:33:01 philips: no chance yet.. sorry, I Mar 24 20:33:31 I'm super busy until ~April3. I really want to squeeze it in one night, but didn't get a chance last night. Mar 24 20:33:41 Keep pinging me, and I'll let you know if I give it a shot. Mar 24 20:33:44 sorry again... Mar 24 20:39:34 snowbike: no worries, just wondering Mar 24 21:03:27 anyone familiar with using TestAnImageInRamUsingRedBoot when the image does not contain a single ramdisk.gz (ref. slugimage) ? Mar 24 21:04:04 seems howto was written when slugtool was still used, and image just contained RedBoot, SysConf, ramdisk.gz and vmlinuz Mar 24 21:10:04 any problems with the leds? Mar 24 21:10:18 wb NAiL Mar 24 21:10:29 not leds, but with module loading Mar 24 21:10:31 read above Mar 24 21:10:59 yeah, I know Mar 24 21:11:02 that has to be solved Mar 24 21:11:26 easy to solve, but not sure about best way to solve it ;) Mar 24 21:13:00 in Debian i would add those modules to /etc/modules Mar 24 21:16:36 Just red there is a usb enclosure compatibility list in http://groups.yahoo.com/group/nslu2-general/database - any reason why this is at yahoo and not public in the wiki? Mar 24 21:16:39 read Mar 24 21:25:11 NAiL: when you test images, do you flash them, or just load them in RAM and exec them? Mar 24 21:25:34 or don't you test :-P Mar 24 21:27:14 flash Mar 24 21:30:47 trying to figure out how the test in ram should work with the FIS directory... it certainly does not work as described in the wiki entry (outdated) Mar 24 22:05:56 looks like it ain't possible, and that rwhitby knows why Mar 24 22:09:54 Is there any project preference for how we handle packages that have a packagename-conf program that tells you how to link with them for cross compiling. Should we modify the copy of *-conf that gets installed in staging to include a reference to the staging dir (specifically optware but I guess the approach for OpenSlug would be the same) Mar 24 23:03:24 03aabaker * 10unslung/make/neon.mk: neon: modify neon-config in staging so it reports correct la file for cross building Mar 24 23:04:30 03aabaker * 10unslung/make/svn.mk: svn: update to version 1.3 Mar 24 23:44:57 03aabaker * 10unslung/Makefile: neon+svn: Mark updated packages as ready for testing Mar 25 00:39:52 how the hell can I run the make menuconfig in order to add support for a few things in the openslug kernel? Mar 25 00:40:21 I have looking over those directory, but I couldn't understand Mar 25 00:42:03 s/have looking/have been looking/ Mar 25 00:42:03 eddyp meant: I have been looking over those directory, but I couldn't understand Mar 25 01:02:32 ARCH=ARM make menuconfig Mar 25 01:13:55 tbm: have you seen this: http://www.nslu2-linux.org/wiki/OpenSlug/ProgrammingTheOpenSlugLEDs ? Mar 25 01:17:35 eddyp: I need to review the LEDS kernel patch first, and RTC Mar 25 01:17:54 I think the new LEDS driver uses /sys Mar 25 04:48:04 * ByronT-Away is now auto-away after 3h idle Mar 25 07:25:16 03bzhou * 10unslung/make/php-fcgi.mk: adding php-fcgi Mar 25 09:33:28 ARE THERE ANY documents on how to make a custom kernel image for opendebianslug? Mar 25 09:51:15 fuck man, is there, at least, a glosary for all the terms used? Mar 25 09:51:44 I don't understand how can one understand your development system Mar 25 09:56:07 eddyp: you have compiled the stock kernel, right? Mar 25 09:56:35 eddyp: 1) be more friendly Mar 25 09:56:43 eddyp: 2) I told you to stay with Debian. :-P Mar 25 09:57:02 then you could at least bitch at me Mar 25 09:58:45 tbm: can you explain me a bit about the FIS directory vs. TestImageInRAM Mar 25 09:59:31 i have been trying to test a slugos-3.6-beta build (debianslug) but fail to use the test in ram procedure Mar 25 09:59:58 i am now pretty certain it does not work, but don't fully understand why not Mar 25 10:01:31 is it true that the kernel is patched to read info from flash, instead of the original ramdisk.gz and that it therefore really need to be in flash? Mar 25 10:02:20 hm, depends Mar 25 10:02:37 i know the d-i works from ram Mar 25 10:02:45 if you have the original firmware, you cannot use the slugos/debian/whatever kernel Mar 25 10:02:55 because the original firmware doesn't use FIS Mar 25 10:03:03 but output when running slugimage is different (valid ramdisk.gz, no Flashdisk) Mar 25 10:03:36 well, I guess the real question is _how_ you did that slugos-3.6-beta build Mar 25 10:03:43 I don't know anything about its build process Mar 25 10:03:51 but are you aware that you have to byteswap the initrd? Mar 25 10:04:00 tbm: does debian need initrd support? Mar 25 10:04:10 NAiL: yes. why? Mar 25 10:04:26 to answer the question: build is done from berlios.de openslug repository Mar 25 10:04:35 because the latest 2.6.16 kernel doesn't have initrd support. Mar 25 10:04:47 NAiL: we need it e.g. to load the usb modules needed to boot from disk Mar 25 10:04:47 so I guess that should be added back in. Mar 25 10:04:47 think keypoint is the slugimage call in the build process Mar 25 10:05:17 NAiL: which latest 2.6.16 kernel? you mean the slugos config? Mar 25 10:05:28 well, we (debian) have our own config file for the kernel Mar 25 10:05:30 slugos config, oe config, cvs config. Mar 25 10:05:39 ah, yeah, makes sense ;) Mar 25 10:06:09 pepijn: if you're trying to use the latest kernel with an initrd, you will fail. Mar 25 10:06:10 tbm: i am running debian, but use the slugos build for booting Mar 25 10:06:32 if that makes sense Mar 25 10:06:40 same thing I do Mar 25 10:07:01 booting debianslug which pivots to disk where I've got a full debian install Mar 25 10:07:06 yep Mar 25 10:07:29 and then we don't need initrd Mar 25 10:07:48 (and no byteswapping) Mar 25 10:07:54 pepijn: well, the problem is that I know nothing at all about slugos, so I've no idea why you wouldn't be able to load it via the ram method Mar 25 10:08:01 you have to talk to slugos people Mar 25 10:08:17 tbm: you were involved in slugimage, weren't you? Mar 25 10:08:34 pepijn: what's slugimage? Mar 25 10:08:42 god, there are too many fricking names... Mar 25 10:08:51 oh, slugimag Mar 25 10:08:53 the tool? Mar 25 10:09:04 slugimage: perl script to produe a flashable image Mar 25 10:09:11 based on slugtool Mar 25 10:09:16 right. no, not really Mar 25 10:09:23 I just made a small patch recently Mar 25 10:09:32 hehe, found your name in some context Mar 25 10:09:53 rwhitby is the man to ak Mar 25 10:10:05 rwhitby: ping? Mar 25 10:10:31 tbm: how do you build the d-i images? that uses a 'real' ramdisk.gz, right? Mar 25 10:10:37 and not a Flashdisk part Mar 25 10:11:00 what's a real ramdisk fs a "flashdisk part"? Mar 25 10:11:14 but yeah, the d-i images use a kernel plus an initrd (ramdisk) Mar 25 10:11:28 well, what problem do you see? Does the kernel not find the initrd? Mar 25 10:11:40 cause I had that problem Mar 25 10:12:45 don't have a serial console, makes me blind ;) Mar 25 10:13:12 but the thing i notice when running slugimage --unpack is that: Mar 25 10:13:30 a) with di-nslu2.img i get a 'valid' ramdisk.gz (can run gunzip on it) Mar 25 10:13:59 you do? Mar 25 10:14:03 I don't Mar 25 10:14:13 let me try Mar 25 10:14:46 let's download http://people.debian.org/~kmuto/d-i/images/daily/nslu2/netboot/di-nslu2.bin and check Mar 25 10:15:43 ok, anyone up for testing the latest and greates leds-script? Mar 25 10:15:43 $ slugimage -u -i di-nslu2.bin Mar 25 10:15:54 b) with a image produced by slugos build i get an 'empty' ramdisk.gz, but also a Flashdisk Mar 25 10:15:55 gunzip ramdisk.gz Mar 25 10:15:58 gunzip: ramdisk.gz: not in gzip format Mar 25 10:16:00 voila Mar 25 10:16:21 pepijn: a Flashdisk... I've no idea what Mar 25 10:16:23 ah sure Mar 25 10:16:24 I know Mar 25 10:16:35 right. Well, Debian and slugos are different Mar 25 10:16:45 Debian uses a real initrd (ramdisk) Mar 25 10:16:57 hmmm... booting my 'science officer' to double check Mar 25 10:17:05 pepijn: OE builds tell the kernel to load direct from Flashdisk partition Mar 25 10:17:13 what slugos does is different: they don't use a ramdisk, but they write this "Flashdisk", which is basically a JFFS2 partition Mar 25 10:17:26 and then they boot the kernel with root=/dev/... (that jffs2 partition) Mar 25 10:17:38 empty ramdisk partition is there cause Linksys redboot loads that into ram, and if you don't put a valid empty on there then it will trash your memory if you are unlucky. Mar 25 10:17:39 and the empty ramdisk is needed because the firmware is fucked Mar 25 10:17:45 you cannot tell it not to load an initrd Mar 25 10:17:48 rwhitby: and therefore the 'test image from ram procedure' is obsolete, right? Mar 25 10:17:51 so you simply give it an empty one Mar 25 10:18:11 right, so the ram procedure won't work in that case Mar 25 10:18:17 it only works for a real initrd Mar 25 10:18:17 pepijn: test from ram is indeed incorrect (as far as the rootfs goes). it's fine for testing new kernels though. Mar 25 10:18:48 when i woke up this morning, i suddenly realised that this (as rwhitby described) was the case ;) Mar 25 10:18:52 pepijn: sorry, I didn't understand the problem inititally Mar 25 10:18:54 just wanted it confirmed Mar 25 10:19:15 pepijn: if you get a chance, please update the wiki to correct the stale information Mar 25 10:19:17 there are so many terms and different ways of doing things... reeally confusing Mar 25 10:19:35 oh yes, i will... if i can find the correct 'wording' Mar 25 10:19:57 rwhitby: speaking of doing things in a different way, even though all of the nslu2 people were against my idea of using APEX I think I'll consider it again Mar 25 10:20:00 tbm: but that's the Linux way, isn't it ;-) Mar 25 10:20:06 rwhitby: can you say something about the status of slugbug ? Mar 25 10:20:08 rwhitby: no, that's the Perl way :-P Mar 25 10:20:28 APEX ? Mar 25 10:20:29 pepijn: it's broke Mar 25 10:20:45 yeah, i meant to file a bug on that :-P Mar 25 10:20:46 pepijn: it's a boot loader... that doesn't have the problems of redboot Mar 25 10:20:53 tbm: as a replacement bootloader or a second stage bootloader? Mar 25 10:21:06 rwhitby: 2nd stage Mar 25 10:21:13 I have no issue with that. Mar 25 10:21:27 rwhitby: given that Debian and slugos/unslug are incompatible anyway, we might just as well go one step further Mar 25 10:21:29 I think it's superfluous, but I have no issue with it :-) Mar 25 10:21:38 NAiL: you got around to looking at the module loading ? Mar 25 10:21:42 and using APEX as a 2nd stage boot loader means we don't have the kernel restrictions Mar 25 10:21:58 rwhitby: superfluous... it would fix the "oh your kernel is too big" problem Mar 25 10:22:00 tbm: size restriction? Mar 25 10:22:01 pepijn: not yet. Been hacking the leds stuff Mar 25 10:22:03 which _is_ getting a problem Mar 25 10:22:12 tbm: if that makes it easer to maintain a debian nslu2 port, then it's a worthwhile step. Mar 25 10:22:27 rwhitby: it would allow me to get rid of the nslu2 specific kernel Mar 25 10:22:34 rwhitby: anyway, I still need to check it out first Mar 25 10:22:40 but i'm leaning towards doing it Mar 25 10:22:48 the apex maintainer is also a Debian guy, so that might be helpful Mar 25 10:23:03 yeah Marc is a good guy Mar 25 10:23:05 and then we could also support the fat slug and that stuff Mar 25 10:23:27 how should we fix the loading of USB support modules in the latest builds ? Mar 25 10:23:27 I basically see no downside, but that's without actually having looked at it in detail Mar 25 10:23:44 but I know about using a boot loader to boot another boot loader to boot the kernel from my work on the Cobalt.. :) Mar 25 10:23:44 pepijn: it should be fixed the way it's supposed to work ;) Mar 25 10:24:15 pepijn: howeve that stuff usually works ;) Mar 25 10:24:16 hehe, how is it supposed to work? which file is/should be responsible for module loading ? Mar 25 10:24:27 /etc/modules (like in debian) Mar 25 10:24:29 ? Mar 25 10:24:40 not entirely sure Mar 25 10:24:54 rwhtiby: opinions? Mar 25 10:25:10 s/rwhtiby/rwhitby/ Mar 25 10:25:10 pepijn meant: rwhitby: opinions? Mar 25 10:25:40 pepijn: module-utils is how it is done in OE. Mar 25 10:25:46 I assumed debian would be the same Mar 25 10:26:23 rwhitby: so the way to fix usb/mod_sd loading is? Mar 25 10:26:43 my problem is not debian related Mar 25 10:26:55 it is slugos and a modularized kernel Mar 25 10:27:13 so modularized that the usb stuff is not *in* the kernel anymore Mar 25 10:27:24 <|kami|> hi is anyone here how runs a working version of Postfix on his slug? Mar 25 10:27:36 so i can boot from flash (not ram, hehe) but the usb stuff is not in place Mar 25 10:27:42 which means /dev/sda1 does not exist Mar 25 10:28:19 we fixed this by changing the RDEPENDS to actually put the usb modules on the rootfs Mar 25 10:28:32 but then i need to modprobe them Mar 25 10:29:00 so issue is: which file needs modification to get the modules loaded before pivotting to /dev/sda1 Mar 25 10:29:47 when rootfs is built, it should put something in /etc/modutils (or some directory named close to that), and call update-modules (or some script named close to that) which then writes /etc/modules (or some file named close to that) Mar 25 10:31:18 my current rootfs only has af_packet.conf in /etc/modutils Mar 25 10:31:45 so how to fix that in the build-system ;) Mar 25 10:33:04 <|kami|> hi is anyone here how runs a working version of Postfix on his slug? Mar 25 10:33:21 10:59 < koen> add a few more module_autoload_ = "module" in the nslu2.conf Mar 25 10:34:09 * NAiL assumes that is an "ok" solution Mar 25 10:34:23 few more? Mar 25 10:34:28 i find none ;) Mar 25 10:34:52 but do find examples for other hardware Mar 25 10:37:02 well, it does look like a valid way to fix the issue Mar 25 10:37:25 yes, trial running it here Mar 25 10:42:29 tbm: the ramdisk.gz from di-nslu2, what i meant was that file gave me some 'sensible' output, not that gunzip worked ;) Mar 25 10:42:40 my bad Mar 25 10:45:53 pepijn: how's the trial coming? ;) Mar 25 10:47:32 hmm Mar 25 10:47:33 this is odd Mar 25 10:47:52 One of my usb-cables doesn't work with my 2.5" hdd Mar 25 10:48:24 at first I thought it was because the drive maxes out at 0.7A... but it does work with another cable. Mar 25 10:48:48 the strange thing is that the "faulty" cable works just fine when I use it elsewhere Mar 25 10:51:16 jordstr?ling? Mar 25 10:51:19 ;) Mar 25 10:52:05 heh Mar 25 10:52:06 dunno Mar 25 10:54:32 *duh* Mar 25 10:54:49 * NAiL screwed up the initscripts **** ENDING LOGGING AT Sat Mar 25 10:59:57 2006